OLPC News 2008-01-19

[Walter Bender]

1. Cambridge: A third Learning Workshop was held at OLPC this past
week. There was excellent attendance and participation; a real network
of laptop learning activists is forming. Workshop attendees are not
merely listening but are contributing to the conceptual basis of
practice in schools and communities. There was a blend between the
conceptual and practical concepts, and the localities beginning will
help innovate the learning environments and communities of the 21st
century. The presentation by Dr. Felton Earls and Maya Carlson of the
Harvard School of Public Health on participatory surveys and
indicators for community development as well as their work in Tanzania
and Chicago was inspirational. The Learning team of Edith Ackermann,
Ed Baafi, Fatimata Seye Sylla, Juliano Bittencourt, Elana Langer,
Julain Daily, Cynthia Solomon, Alice Cavallo, and David Cavallo
contributed mightily. Special thanks for support especially to Felice
Gardner, as well as Tracy Price and Jennifer Amaya. As always, a
highlight is the Activity Open House where developers demonstrate
their activities on the XO.

2. G1G1: During the reconciliation process of the "get" laptops
shipped during Give One Get One, a number of unfulfilled order records
were uncovered. The OLPC team has been working hard with our partners
to resolve all open issues. We expect another ~5000 XO laptops will be
shipped on Monday. The remaining orders pose an extra challenge as
they either have incomplete or no shipping and contact information. If
you have not yet received your XO laptop, you should be getting an
individualized email that addresses your specific situation. If you
are scheduled to receive your laptop next week, you will also be
getting a follow-up email with tracking information. We'll be adding
additional phone lines and shifting agents to reduce wait times. A
further reconciliation of the data will be conducted this week,
although hopeful, we can anticipate additional incomplete orders. Our
apologies for these delays.

3. Ulaan Baatar: Enkhmunkh Zurgaanjin, Carla Gomez Monroy, Jan
Jungclaus, and RedHat's David Woodhouse are working hard to set up a
structure that can provide sustainability to the project in Mongolia
such that it can spread it throughout the country. On Wednesday, the
Minister of Education visited the school for the "laptop hand out"
event. On Friday, an optical-fiber cable was set up, in spite of the
extreme low temperatures; on Saturday, the schools were connected to
the Internet. David has been working with a group of local technical
people on the servers and Internet set up infrastructure as well as on
configuration. John Watlington has been providing support remotely
from OLPC.

We have been meeting almost every evening with the strategic team of
the Ministry of Education to provide feedback and sort out challenges.
We met yesterday with the Ministry of Education team, teachers,
principals, ICTA, content team and pilot research team to provide
detailed feedback of how the project is going so far and to bring up
things to be considered for the short and long terms.

Teachers are putting their hearts into the program. They had their
first sessions with the children. Parents, too, have shown support.
And the children, of course, love it. The Constructionist model of
learning has found wide-spread support within the MoE.

There are more photos (See http://wiki.laptop.org/go/Ulaanbaatar) in our wiki.

4. School Server: John Watington reports that we have a new build
which supports schools with multiple servers in a school and including
a Jabber server! Build 150 was released, along with lengthy
configuration notes (See
http://wiki.laptop.org/go/XS_Installing_Software#OLPC_XS_150).

The configuration interface is still stone knife and bear skin, but
functionality appears to be there. We hope to have a build improving
the configuration process and adding web caching by the end of next
week.

David Woodhouse is in Ulaan Baatar, Mongolia helping Carla Gomez
Monroy deploy school servers along with the laptops. The servers we
shipped from Cambridge have arrived and are being installed. David has
been handling the difficult task of positioning two servers (with six
antennae) to cover a three-floor school. He is also facing the need to
upgrade the laptops right away to avoid a networking meltdown. The
good news is that the school is finally connected to the Internet; we
can assist from Cambridge.

5: Firmware: Richard Smith fixed the "repeated game keys on resume
bug" (Ticket #2401). During a resume, the main processor is not ready
to receive key codes for about 100ms after the delivery of the SCI
wakeup event. The EC dealt with this long delay badly. Fixing this
should unblock ebook mode. Richard released a new EC code version that
is available in Firmware Q2D09. This should show up in a signed
Joyride build soon.

6. Battery issues: Richard did a large amount work on batman.fth; he
added the ability to run a manual charge while watching voltage,
current, and accumulated current registers. He received three laptops
from the field with battery problems. One has a battery that just
won't take a charge. The other two laptops won't recognize that the
battery is present. Richard plans on tearing into these two machines
next week to determine root cause. We need to discuss with Quanta what
to do with problem batteries.

7. Schedules: Final testing for Ship2.2, Build 656, was held up this
week as we tried to iron out some final details of the new process for
"Unscheduled Software Release" (USR) (For those interested in process,
please see  http://wiki.laptop.org/go/Unscheduled_software_release_process).

Update 1 is about to start the release process. Highlights include:
• suspend and resume is mostly, but not entirely implemented;
• Rainbow security is enabled;
• the Browse activity has been updated to use the technology in
Firefox 3 Beta 2, which is significantly faster and better at memory
usage. Innumerable bugs have been fixed. Performance is significantly
improved. Memory and file leakage greatly improved.

Update.1RC1 (release candidate 1) is almost ready (Monday with luck).
Joyride 1551 is very, very close to Update.1's contents, although it
has some additional activities bundled in the base system that we do
no plan to ship in Update.1. Please do test it, along with the new
firmware version (Q2D09).

We will then start a testing cycle:
• suspend-and-resume cycling for reliability—during the run up to mass
production the laptop was found to be able to reliably suspend and
resume at least 50,000 cycles (the length of the tests we were
willing to tolerate). We need to ensure there are not software or
firmware regressions in this area.
• scaling tests—we need to ensure sane behavior of the systems in
circumstances such as 300 children resuming their laptops all at once
in the morning.
• verification of power use in different use states, on our power
measurement systems;
• wireless driver testing and upgrade testing;
• testing with the school-server software;

RC2 (release candidate 2) will pick up additional translations and key
bug fixes that missed RC1. The community is working to complete the
Spanish translation, which was not complete by the string freeze date
due to the holidays. There will therefore be a refresh of packages to
complete the translations. Activity developers should only be picking
up translations (and fixes for approved bugs).

RC3 (release candidate 3) is intended to pick up critical bug fixes
discovered during testing, and is the first candidate that is a real
candidate for widespread release.

Joyride will be reopened for the start of Update.2 development later
this week.

8. Testing: Chih-yu Chao has created the Update.1 Test page, which
outlines the major features and bug fixes of this release and links to
the test cases to be run and their results (See
http://wiki.laptop.org/go/Update.1). The goal is to have enough
planning and information around this release to be able to ask for
help from the development and test community over the next two weeks
to really hammer on the release. If we can do it, this will be the
first release that gets organized, methodical community testing. Keep
tuned to your email for more information (testing, devel, sugar
mailing lists).

Yani Galanis is back from a short hiatus and has jumped right back
into various wireless testing and debug activities. So far he is happy
with David Woodhouse's rewrite of the wirless driver. He has fixed
some problems with olpc-netstatus so it will accurate report the
laptops network and mesh configurations. He was able to get
olpc-netlog working again (with Noah Kantrowitz's help) to zip up the
logs, and olpc-netcapture to capture network traffic.

9. Support: This week, when all the laptops to Give One Get One donors
were anticipated to have been delivered, Kim Quirk  suggested we ask
people to send email if they are still waiting for their laptop. When
we hit 100 emails in less than a day, it became obvious that were
dealing with a much larger problem than anyone at OLPC had imagined.
This prompted some quick meetings between the companies involved in
the order processing and distribution to try to get a handle on the
scope of the problem and how to fix it fast. We learned about orders
that could not be matched up between daily and monthly reports, orders
that do not have enough information to ship, addresses that couldn't
be verified, PO boxes, and miscellaneous other issues—about 10% of the
total order volume. Adam Holt's support gang (up to 55 people now!)
were inundated by mid-week with donor information requests, as was the
Donor Services 800 number. Adam recruited two of the volunteers, Sandy
Culver and Steve Holton, to join Greg Babbin and Adam to access the
shipping database to help answer these requests.

Thanks to the entire support team who have been working day and night
to respond to these extra requests!

10. Datastore: Ivan Krstić ran a Journal/datastore summit at OLPC this
week. In attendance through out the week were Marco Pesenti Gritti,
Tomeu Vizoso, Eben Eliason, Erik Blankinship, and Bert Frudenberg. A
number of other members of the core team and the community joined
periodically. It was a very productive week: the team nailed down
almost all the details required before a first pass at implementation
can begin. But before we do so, and while we continue conversations
about the new API, Ivan will publish specification in the next few
weeks for a round of public discussion. Look forward to a new object
model, a refined set of interactions, and new features such as
versioning and action-based journal entries.

11. Sugar activities: Arjun Sarwal incorporated sensor input into
Turtle Art this week. One can control any aspect of the Turtle's
motion based on sensor input. The next step is to integrate the
concept into a Turtle Art "block" (See
http://wiki.laptop.org/go/Measure#Sensor_Input_into_Turtle_Art).

Arjan has been talking to educators and teachers how they can organize
some activities around the Measure Activity. He has also spoke with
representatives of the Illinois Math and Science Academy (IMSA)
chapter who have made a video documentary of experiments with sound
using the Measure Activity (See
http://wiki.laptop.org/go/Illinois_Math_and_Science_Academy_Chapter).

Manusheel Gupta is investigating options for building a spreadsheet
actvitiy for the XO. Python-powered spreadsheet (PPSS) seems to be a
good choice for integrating into the Sugar environment, while perhaps
pulling in some features from GNumeric. Eben Eliason will be
discussing the ideas on the UI of the spreadsheet during the coming
week. (See http://olivier.friard.free.fr/software/ppss/index.php).

Simon Schampijer fixed an error in the download handling within the
Browse activity (Ticket #6018). Dan Williams and Simon finally think
we have a good solution for "airplane mode", e.g., operation with the
radio off. A new network manager went into Joyride-1548 and the sugar
rpm is building.

Simon does not really understand what happens in regard to reports
that Browse is running slow after an update to Update.1 (Ticket #6046)
(as opposed to a clean install).

Eva Schroth successfully conducted an interview using the XO laptop's
Record activity: after modifying some constants, she was able to
record an one hour conversation.

On the Etoys front, most of the core team members visited Poitiers,
France this week; an IEEE conference called C5 was held. Many
researchers, educators and Squeakers who are interested in
collaboration and education met together and had interesting
conversation. Bert, in response to Arjun's Turtle Art demonstration,
exposed some code in Etoys to enable the microphone level to be used
as a data stream within scripts.

12. System software medley: Giannis Galanis contributed network fixes
to olpc-utils that solve two Update.1 blockers. Phil Bordelon sent a
tool to cleanup orphan Journal previews, which was also an Update.1
blocker.

FFM packaged up the python-gasp, an API wrapper for pygame for new
programmers, which has just gone through the Fedora review process.

Chris Ball worked on OHM timing code, and with Reinier Heeres on
fixing "ebook mode" to work inside Rainbow.

Reinier Heeres mostly worked on improving and build-testing of some
bugfixes of last week. He also activated the new build announcer
script, which needed a few minor fixes. For Sugar he fixed an issue
with the stop button disappearing when rotating the screen (#5824),
and for Read the eBook suspend problem (#1396).

Andres Salomon mostly worked on the touchpad driver this week and has
made great progress on fixing problems, and improving its behavior.
More importantly, Andres work made clear we should use the tablet
sensor in relative mode by default, a conceptual breakthrough that had
eluded us. A test kernel with the new driver is available here:
http://dev.laptop.org/~dilinger/master/kernel-latest.i586.rpm.

13. Presence: Dafydd Harries spent most of this week again working on
the Jabber server component. A large part of this was working out how
to effectively expose the person/activity information that the
component will store over XMPP. Dafydd thinks he has a reasonable
protocol; he plans to set up a test server that we can measure
performance against.

Robert McQueen attended an introductory conference call with Ivan and
Jon Herzog about collaboration and security; they are planning a
specification-writing fest in late February.

Morgan Collett has been going through the wiki getting the references
to Presence, Telepathy and Tubes up to date, and working on a more
comprehensive reference for Presence Service. Morgan also modified
Chat to make sure web links copied to the clipboard can be pasted in
Write, Web location bar, and Terminal (Ticket #6066). That patch will
land when the Spanish translation of Chat is complete.

Guillaume Desmottes continued work on Hyperactivity, a collaboration
stress testing tools (Ticket #5817). It is now able to
create/join/leave activities, set up and use D-Bus tubes. Guillaume
started to test Salut using it and discovered some interesting bugs;
most of them are already fixed.

14. Localization: Bernie Innocenti has been doing some integration
work with Manusheel Gupta on Devanagari input support, but it seems
there's more work to do, especially in the Write activity. Bernie met
with Lidet Tilahun for a roundup on our Ethiopian support, and filed a
bunch of bugs out of it. Lidet will contribute translations in Pootle.

Sayamindu Dasgupta reports that we have new teams for Dari, Fula and
Telugu. He also tracked down a problem in Pootle that was preventing
him from updating the PO files in the XO Bundled project. This has
been quite difficult to trace down. The rest of the week was spent on
more mundane things:
• he polished and debugged the various helper scripts that is used to
run Pootle more smoothly;
• he helped Simon cross check the list of languages that are given as
options by the sugar-control-panel (In the process, they identified a
few languages that would require new locales to be added to glibc in
order to be supported);
• he helped a number of users get started with the translations; and
• he added Slider Puzzle to Pootle.

Dr. Habib Khan reports that localization into Pashto is in final phase
and that after some confusion on the Pootle server in regard to Dari
and Farsi, progress in being made there as well.

15. Build system: Dennis Gilmore submitted patches to rpm enabling
support for the AMD Geode. He has done some work on koji in
preparation for supporting us. Patches will be submitted next week for
upstream inclusion. These add Geode support and allowing us to pull
upstream builds into our instance. Once initial support is in koji,
Dennis want to add support to allow .xo building. This would result in
a side effect that we get a .src.rpm and .noarch.rpm out of the
process

We will need to have a git tree setup that will mimic Fedora's cvs for
things that we keep out of Fedora.

Dennis and Michael Stone looked at possibly using livecd-tools for
Update.2: what would be involved in it and if its worth the effort.

This week, Michael talked with Bernie, Scott, and Dennis on ways to
improve
build infrastructure, offered occasional questions in the Journal
summit, and diagnosed the 'upgrade-server can't download builds' bug.

Update.1 is mostly synced up with Joyride. There are a few small
pieces that need to be finished. As noted, we are very close to having
an Update.1

16. Content: The inclusion of the Doom activity in the wiki has
sparked a healthy email discussion about content and filtering.
Although heated at times, it has generally been productive. The gist
of the debate revolves around the twin issues of (1) should OLPC be
adjudicating what is appropriate content and (2) how should content be
tagged such that children, parents, teachers, and others can make
informed decisions about what content they access.

Suggestion have ranged from adopting "Terms of Use" such as those
found on the Scratch website (http://scratch.mit.edu/terms) to
fleshing out our guidelines
(http://wiki.laptop.org/go/Activity_guidelines) to making it easier
for community members to search and sort favorites (requiring possible
extentions to MediaWiki).

This discussion is by no means over, but please continue the thread on
the olpc-open <olpc-open at lists.laptop.org> list rather than devel,
which is intended for discussion of technical rather than policy
topics.

17. OLPC Health: Arjun Sarwal continues his efforts to organize the
community in medical and health applications around the XO laptop. He
reports that we have a growing list of volunteers in three areas:
(1) Creating a Library/repository of information that would be shipped
on the XO laptop as part of the default software on it. This would be
a ready reference for preliminary diagnosis of diseases and a
reference for symptoms. This would also include general information on
an array of topics such as hygiene, nutrition, balanced diets, etc.
(2) Developing software that asks the user a series of questions and
helps in a preliminary diagnosis. Links to useful websites and online
portals.
(3) Developing and using hardware peripherals that connect to the XO
laptop. These include, but are not limited to the build-in camera
(with the possibility of add-on optical elements; an EKG; and a pulse
oxymeter.

18. Activity Handbook: Christoph Derndorfer reports that the first few
chapters of an Activity Handbook are finished. The purpose of this
handbook is to provide all the information needed in order to get
started with software development for the OLPC XO. The current draft
includes the first four chapters:
1. Welcome to the Activity Handbook!
2. Introduction to Sugar
3. Preparation
4. Sugar Basics

Christoph et alia  will be expanding the handbook over the coming
weeks to include chapters about using the Journal, collaboration,
using the various XO input devices and "Sugarizing" software. (Please
see
http://www.olpcaustria.org/mediawiki/index.php/Activity_handbook and
http://www.olpcaustria.org/mediawiki/upload/a/af/Handbook_20080113.pdf).

19. Hello World: In a related effort, Chris Hager and Jaume Nualart
report that they have created two new tutorials (during a
"pizza-and-beer" coding session) for creating Activities with PyGTK,
one of them using Glade (See
http://wiki.laptop.org/go/PyGTK/Hello_World_Tutorial).
Chris and Jaume are using activity.py as a wrapper, which loads the
code and GTK interface from gtktest.py. This way, very little code is
required to get a PyGTK Activity running in Sugar—just six lines in
gtktest.py—and PyGTK Activities can run as standalone versions on any
Linux system by default.

Example Bundles:
http://wiki.laptop.org/images/b/ba/Gtktest.xo
http://wiki.laptop.org/images/0/02/Gtktest-glade.xo

20. Mongolia: Dave Woodhouse is in Mongolia setting up servers in two
schools, which as been an educational experience. Firstly, the
wireless penetration through the walls they have here to cope with
temperatures of –40°C is fairly dismal—Dave reports that we are having
to use a lot of active antennae to get the coverage we need. We're
laying them out as if they were "normal" access points, to try to get
coverage of all the rooms they'll be teaching the 2nd–5th grades in.
Hopefully, the nature of the mesh will improve coverage.

To start with, each school will have five antennae, with two servers.
That setup will be re-evaluated when it's fully deployed and tested in
the classrooms. It is physically installed in one school so far, and
fully
cabled (including CAT5 to the other rooms where they have computers).
The other school should be similarly set up by the end of Monday.

21. Pakistan: Habib reports progress on the e-book project in
Islamabad. Eight elementary text books based on curriculum of the
Federal Ministry of Education, Islamabad have been made into e-text
books.

-walter

-- 
Walter Bender
One Laptop per Child
http://laptop.org